Kusnadi, Adhi (2017) Security system with 3 dimensional face recognition using PCA method and neural networks algorithm. In: 2017 4th International Conference on New Media Studies (CONMEDIA), 8-10 Nov. 2017, Yogyakarta.
Full text not available from this repository.Abstract
The increasing use of computers and the internet has resulted in an increasing trend of computer crimes. This is the reason computer security is becoming important. A computer security system should be able to provide some kind of information assurance such as confidentiality, integrity, availability, authenticity and non-repudiation of data. The method used to help meet the authenticity of information assurance is biometric-based authentication, among others, a two-dimensional (2D) face recognition system, but it can make mistakes in recognition. The intruder is also easier to enter the system if has a photo print from the user's face. To overcome these deficiencies can be use a three-dimensional (3D) face recognition system. This research did three-dimensional (3D) face recognition by not doing 3D face reconstruction. But using face data got from camera ToF i.e. distance, amplitude, and intensity from each image pixel as input data. The hypothesis was face recognition execution speed faster and similar accuracy when compared with research conducted by Zhang and Lu. The algorithm used in this research, is back propagation neural networks algorithm and Principal Component Analysis (PCA). Obtained accuracy of 95% and training time of 9728 seconds. Face recognition in this study has a lower accuracy level than previous research but faster face recognition speed.
| Item Type: | Conference or Workshop Item (Paper) |
|---|---|
| Subjects: | 000 Computer Science, Information and General Works > 000 Computer Science, Knowledge and Systems > 006 Special Computer Methods |
| Divisions: | Faculty of Engineering & Informatics > Informatics |
| Depositing User: | Administrator UMN Library |
| Date Deposited: | 06 Oct 2021 08:19 |
| Last Modified: | 02 May 2023 08:49 |
| URI: | https://kc.umn.ac.id/id/eprint/18558 |
Actions (login required)
 |
View Item |
